JP2008527555A - Method, apparatus and program storage device for providing automatic performance optimization of virtualized storage allocation within a virtualized storage subsystem - Google Patents

Abstract

  A volume provisioning advisor for managing storage allocation in multiple virtual storage subsystems is disclosed. Requests for data storage and workload requests are received from users. Performance information is received for the storage controller and management disk in the virtualization engine. Resource groups are defined based on received performance information and available managed disks, and their usage tracked. Resource group performance within the virtualization engine is predicted based on user workload demand and current or historical resource group utilization. Storage is allocated to one or more managed disks in the virtualization engine based on predicted resource group performance.

Description

  The present invention relates to network storage systems, and more particularly to a method, apparatus and program storage device for providing automatic performance optimization of virtualized storage allocation within a virtualized storage subsystem.

  In enterprise data processing mechanisms such as those used in companies, government agencies, or other business entities, information is often stored on a server and accessed by a user over a network, for example. The information may include any type of program and / or data to be processed. Users using their personal computers, workstations, etc. (generally “computers”) can allow their computers to obtain information to be processed and store the information on, for example, a remote server. To.

  In general, a server stores data in a mass storage subsystem that typically includes many disk storage units. Data is stored in units like files. In one server, one file may be stored on one disk storage unit, or several parts of one file may be stored on several disk storage units. Also good. The server can service access requests from many users simultaneously, and the access operations that are serviced simultaneously relate to information distributed across multiple disk storage units so that those access operations are serviced simultaneously. It will be appreciated that this is preferred. In other words, in a way that other disk drive units are lightly loaded or idle, while one disk drive unit is overloaded, i.e. not busy servicing access. It is generally desirable to store information in a disk storage unit.

  A business computer network may have multiple storage networks that are located at a distance from each other and from business users. The storage network may be hosted on different types of systems. In order to perform the work correctly, business users may require quick and reliable access to data contained in all of the storage network. Information science (IT) personnel must be able to provide business users with fast and reliable access.

  A storage area network (SAN) is a high-speed, high-bandwidth storage network that logically connects data storage devices to servers. In turn, business users are typically connected to the data storage device via a server. SAN extends the concepts presented by traditional server / storage connections to provide greater flexibility, availability, integrated management and performance. SAN is the first IT solution to enable users to access information in the enterprise at any time. In general, a SAN includes management software for defining network devices, such as hosts, interconnect devices, storage devices, and network attach server (NAS) devices. SAN management software also allows links to be defined between devices. Within a SAN, software can be characterized as a single virtual disk while defining virtual storage where data is stored across multiple storage disks.

  One key component in reaching this goal of providing business users with fast and reliable access is to ensure that the SAN is fully understood by those who design and maintain the SAN. is there. SANs are often difficult to understand quickly due to their complexity. Tools that allow the SAN and the configuration of virtual systems within the SAN to be quickly understood and changed are beneficial.

  One benefit of SAN is the elimination of bottlenecks that can occur in servers that manage storage access for a large number of clients. By allowing shared access to storage, the SAN can be prepared for lower data access latency and improved performance. However, in large storage networks such as SAN-attached storage, it is not possible for the storage administrator to know where to allocate the increase in storage so that the newly allocated space will achieve the best possible performance. Have difficulty. This is because of the complexity of the network that can include many virtualized storage subsystems, the complexity of analyzing the workload, and the physical storage attributes can be hidden from the application.

  In the past, storage allocation for large storage environments has been performed manually. Storage management software is available that can allocate storage or recommend where to allocate based on algorithms. Nevertheless, these algorithms do not actually attempt to meet production performance requirements within the constraints of available storage including virtual storage systems within the SAN.

  The present invention discloses a method, apparatus and program storage device for providing automatic performance optimization of virtualized storage allocation within a virtualized storage subsystem.

  According to a first aspect, a program storage device is provided, the program storage device comprising program instructions executable by a processing device that performs operations for managing storage allocation in a virtual storage system. The operations include the steps of evaluating the workload profile, determining the performance characteristics of the managed disk, determining the relationship between the managed disk and the resource group, and the resource group to which the managed disk is assigned. Based on, recommending allocation of at least one virtual disk comprising a set of managed disks.

  In one embodiment, the operation further includes defining a workload profile.

  In one embodiment, the operation further includes creating a virtual disk that includes a set of managed disks based on the resource group to which the managed disk is assigned.

  In one embodiment, determining the relationship is based on user input.

  In one embodiment, determining the relationship is based on automatically provided input.

  In one embodiment, creating a virtual disk further includes creating a plurality of virtual disks in a set of managed disks based on a resource group to which the managed disk is assigned.

  In one embodiment, the location for a virtual disk is selected by identifying a set of performance characteristics associated with a set of managed disks and a resource group that includes throughput capabilities for a particular workload. The resource utilization is estimated based on a set of performance characteristics associated with the managed disk and resource group.

  In one embodiment, defining the workload profile comprises selecting at least one selected from the group consisting of a desired amount of storage, a workload characteristic identification and desired throughput level, and an allocation hint. In addition.

  In one embodiment, managed disk performance capabilities are obtained.

  In one embodiment, obtaining the managed disk performance capability further comprises providing manual input indicating the performance capability of the managed disk.

  In one embodiment, obtaining the performance capability of the managed disk further includes receiving the performance capability from the managed disk configurator.

  In one embodiment, receiving the performance capability from the managed disk configurator further comprises obtaining information about the performance characteristics of the managed disk and the associated resource group of the managed disk, the resource group comprising: Identify potential relationships between disks.

  In one embodiment, obtaining the performance capability of the managed disk further includes performing a controlled calibration and using the results from the controlled calibration.

  In one embodiment, performing the controlled calibration includes operating a specified I / O load for the management disk to identify the behavior of the management disk.

  In one embodiment, obtaining the performance capability of the managed disk further includes performing an uncontrolled calibration and using results from the uncontrolled calibration.

  In one embodiment, performing uncontrolled calibration includes analyzing performance behavior as the application operates as an uncontrolled workload to identify managed disk behavior.

  In one embodiment, obtaining managed disk performance capabilities further comprises receiving input from a user via an interface.

  In one embodiment, determining the performance characteristics of the management disk further includes identifying a resource group to which the management disk belongs.

  According to one embodiment, a program storage device is provided, the program storage device comprising program instructions executable by a processing device that performs operations for managing storage allocation in a virtual storage system. The operations include defining the workload profile, determining the performance characteristics of the managed disk, determining the relationship between the managed disk and the resource group, and the resource group to which the managed disk is assigned. Creating a virtual disk that includes a set of management disks based on.

  According to one embodiment, a device for managing storage allocation in a virtual storage system is provided. The device includes a memory for storing storage system information and a processor coupled to the memory. The processor determines the user interface used to define the workload profile as well as the performance characteristics of the managed disk, and the relationship between the managed disk and the resource group is based on user-defined or automated inputs. And is configured to provide a virtual disk allocator for creating a virtual disk that includes a set of managed disks in view of the resource group to which the managed disk is assigned.

  In one embodiment, the processor creates one or more virtual disks in the set of managed disks taking into account the resource group to which the managed disk is assigned.

  In one embodiment, the processor allocator identifies a set of performance characteristics associated with a resource group that includes a set of managed disks and a throughput capability for a particular workload, and the set of managed disks and the resource group Recommend locations for virtual disks by estimating optimal resource utilization based on performance characteristics associated with.

  In one embodiment, the workload profile includes at least one selected from the group consisting of an amount of desired storage, identification of workload characteristics and a desired level of throughput, and allocation hints.

  In one embodiment, the processor provides a graphic interface.

  In one embodiment, the processor provides a scripted application that uses a command line interface.

  In one embodiment, the processor provides a scripted application using an application programming interface (API).

  In one embodiment, the processor adjusts the performance capability of the managed disk using at least one of the group comprising manual inputs, managed disk configurators, controlled calibration results, uncontrolled calibration results and standard interface inputs. decide.

  In one embodiment, the manual input includes selection from a managed disk performance profile.

  In one embodiment, manual input includes selection from a performance profile resource group.

  In one embodiment of the invention, a method is provided for managing storage allocation in a virtual storage system. The method is based on user-defined or automated inputs for obtaining a user-defined workload profile, determining the performance characteristics of the managed disk, and the relationship between the managed disk and the resource group. And determining a virtual disk including a set of managed disks in consideration of a resource group to which the managed disk is allocated.

  According to another aspect, evaluating a workload profile, determining a performance characteristic of the management disk, determining a relationship between the management disk and the resource group, and a resource to which the management disk is allocated Recommending the allocation of one or more virtual disks comprising a set of managed disks based on a group.

  According to one embodiment, a volume provisioning advisor for managing storage allocation in a virtual storage system is provided. The Volume Provisioning Advisor includes a user interface that is used to define workload profiles and a virtual disk allocator that is operatively coupled to this user interface, and the virtual disk allocator is a performance characteristic of the managed disk. A virtual disk containing a set of managed disks, taking into account the resource group to which the managed disks are assigned, determining the relationship between managed disks and resource groups based on user-defined or automated inputs Create a disc.

  In one embodiment, the workload profile includes at least one selected from the group consisting of an amount of desired storage, identification of workload characteristics and a desired level of throughput, and allocation hints.

  In one embodiment, the resource group includes a name, performance attributes, and a set of managed disks that belong to the resource group.

  In one embodiment, the managed disk belongs to one or more resource groups, and the resource group identifies potential relationships between managed disks.

  According to another aspect, a volume provisioning advisor is provided for managing storage allocation in a virtual storage system, the volume provisioning advisor comprising a workload evaluator and a virtual disk disk that evaluates a workload profile. Including the allocator, the virtual disk allocator determines the performance characteristics of the managed disk, determines the relationship between the managed disk and the resource group, and determines the set of managed disks based on the resource group to which the managed disk is assigned It is recommended that one or more virtual disks be included.

  According to another embodiment, a volume provisioning advisor for managing storage allocation in a virtual storage system is provided, the volume provisioning advisor comprising means for obtaining a user-defined workload profile; Means for determining performance characteristics of the management disk and means for recommending allocation of a virtual disk including a set of management disks in view of a resource group to which the management disk is allocated.

  In another embodiment of the present invention, another volume provisioning advisor is provided. This embodiment of the Volume Provisioning Advisor provides a means for obtaining user-defined workload profiles, a means for determining performance characteristics of managed disks, and a relationship between managed disks and resource groups. Means for determining based on user-defined or automated inputs, and means for creating a virtual disk including a set of managed disks in view of a resource group to which the managed disk is assigned.

  According to another embodiment, another volume provisioning advisor is provided. This volume provisioning advisor takes into account the means for obtaining a user-defined workload profile, the means for determining the performance characteristics of the managed disk, and the resource group to which the managed disk is allocated. Means for creating a virtual disk containing the set.

  According to another embodiment, a method is provided for managing storage allocation in a virtual storage system, the method comprising obtaining a user-defined workload profile and determining performance characteristics of a managed disk. And recommending allocation of a virtual disk including a set of management disks in consideration of a resource group to which the management disk is allocated.

  In one embodiment, a virtual disk that includes a set of managed disks is created taking into account the resource group to which the managed disk is assigned.

  In one embodiment, creating a virtual disk comprises identifying a set of performance characteristics associated with a resource group that includes a set of managed disks and a throughput capability for a particular workload; Estimating resource utilization derived from the set of identified performance characteristics associated with the set and resource group.

  In one embodiment, obtaining a user-defined workload profile comprises selecting at least one from a group consisting of an amount of desired storage, identification of workload characteristics and a desired level of throughput, and allocation hints. Further included.

  In one embodiment, determining the performance capability of the managed disk comprises at least one of the group consisting of manual input, resulting configuration data, controlled calibration results, uncontrolled calibration results and standard interface inputs. Using a step.

  Embodiments of the present invention will be described by way of example only and with reference to the following drawings, in which like reference numerals represent corresponding parts throughout.

  In the following description of the embodiments, reference is made to the accompanying drawings that form a part hereof, and in which is shown by way of illustration specific embodiments in which the invention may be practiced. It should be understood that other embodiments may be utilized. This is because structural changes can be made without departing from the scope of the present invention.

  The present invention provides a method, apparatus and program storage device for providing automatic performance optimization of virtualized storage allocation within a virtualized storage subsystem.

  FIG. 1 illustrates a computer network 100 in the form of a local area network (LAN). In FIG. 1, a workstation node 102 is coupled to a server 120 via a LAN interconnect 104. Data storage 130 is coupled to server 120 via data bus 150. The LAN interconnect 100 may be any number of network topologies, such as Ethernet.

  The network shown in FIG. 1 is known as a network client-server model. A client is a device connected to a network that shares services or other resources. Server 120 manages these services or resources. The server 120 is a computer or a software program, and provides a service to the client 102. Services that can be managed by the server include access to the data storage 130, applications provided from the server 120 or other connected nodes (not shown), or sharing of the printer 160.

  In FIG. 1, a workstation 102 is a client of a server 120 and shares access to a data storage 130 managed by the server 120. If one of the workstations 102 requires access to data storage 130, workstation 102 submits a request to server 120 via LAN interconnect 100. Server 120 responds to requests for access from workstation 102 to data storage 130. Possible interconnection technologies between servers and storage are Fiber Channel Protocol (FCP) or Small Computer System Interface (SCSI).

  As the network as shown in FIG. 1 grows, new clients 102 can be added, more storage 130 can be added, and service requests can increase. As described above, the server 120 responds to all requests for access to the storage 130. As a result, the IO load on the storage 130 may increase dramatically, and the workload on the server 120 may increase dramatically, resulting in potential performance degradation. Storage area networks (SANs) have become increasingly popular in recent years to help reduce the bandwidth limitations of traditional client-server models. Storage area networks interconnect with servers and storage at high speed. By combining existing networking models such as LANs with storage area networks, the performance of the entire computer network can be improved.

  FIG. 2 illustrates one embodiment of a SAN 200 according to an embodiment of the present invention. In FIG. 2, server 202 is coupled to data storage device 230 via SAN interconnect 204. Each server 202 and each storage device 230 is coupled to a SAN interconnect 204. Server 202 can directly access any of the storage devices 230 connected to the SAN interconnect. The SAN interconnect 204 can be a high speed interconnect, such as Fiber Channel or SCSI. As FIG. 2 shows, the server 202 and the storage device 230 include networks within and outside of themselves.

  In the SAN 200 of FIG. 2, no server 202 is dedicated to a particular storage device 230 as in the LAN. Any server 202 can access any storage device 230 on the SAN 200 in FIG. Typical characteristics of the SAN 200 may include high bandwidth, multiple paths from the server to storage nodes, large connection distances, and very large storage capacity. As a result, with the complexity of SANs, the performance, flexibility and scalability of Fiber Channel based SAN 200 can be significantly greater than that of typical SCSI based systems.

  FIG. 2 also shows a network administrator 270 coupled to the SAN interconnect 204. It is particularly important to allocate storage 230 in SAN 200 in a manner that effectively utilizes all available SAN resources, from disk to data path, and provides sufficient data protection and resiliency. Administrator 270 may be configured to assist in selecting a storage location within a large network of storage elements. The administrator 270 includes a virtual disk allocator (VDA) 272 that, according to embodiments of the present invention, may include a desired level of performance, user workload attributes, storage varying performance attributes, and different types of Given its response to the workload, as well as the presence of competing workloads in the network, it processes input / output storage allocation according to customer specified performance and space requirements.

  The virtual disk allocator 272 satisfies the storage element's request for storage in the network in a manner that meets the performance requirements specified by the request or via the storage policy mechanism. The virtual disk allocator 272 may operate in an environment such as the IBM® 2145 SAN Volume Controller (SAN VC), which is a virtualized storage subsystem. (IBM is a registered trademark of International Business Machines Corporation.) Virtual disk allocator 272 determines the performance characteristics of the managed disk. The virtual disk allocator 272 determines the relationship between the managed disk and the resource group based on user-defined or automated inputs and determines the resource group and the cache and data path to which the managed disk is assigned. A virtual disk including a set of managed disks is created in consideration of such resource group storage resources.

  Virtual disk allocator 272 extends policy-based thinking to the oven system environment and automates the selection of storage elements within the virtualized storage subsystem to meet performance requirements. Supports the concept of striped volumes such as OS or volume manager software, or DB2® and other database products when selecting and presenting storage elements within a virtualized storage system ( Consider optimal use of applications (such as database applications). (DB2 is a registered trademark of International Business Machines Corporation.) Virtual disk allocator 272 also extends the idea of allocating storage in view of long-term data usage patterns. Virtual disk allocator 272 implements various algorithms required to make intelligent selection of data placement.

  The virtual disk allocator 272 determines which nodes, ie, engines such as the virtualization engine 274, may access the data, and which managed disk group (MDG), ie the group of disks that support the virtual disk, A determination of whether to configure a LUN to be selected may be made. There is at least one managed disk in the MDG, which is used by the virtualization engine 274 and volume manager 276 to stripe the data in the virtual disk, which is an entablized storage system. This corresponds to a logical disk of (ESS). Thus, the virtual disk allocator 272 can select multiple LUNs in multiple resource groups across a single LUN or multiple storage elements to meet a customer's desired performance level.

  The administrator can perform a calibration process 278 to find the performance capabilities of the disks connected to the system. This does not simply use specific knowledge about the performance capabilities of individual disks organized in a specific configuration in a resource group, but rather to find out the performance parameters of those disk groups. Of course, testing will also be included.

  FIG. 3 illustrates an attribute table 300 implemented in the virtual disk allocator 272 according to an embodiment of the present invention. These attributes include user workload attributes and desired performance level information (310), determination of performance characteristics of managed disks or any resource group (312), user-defined or automated inputs. A virtual disk (316) having a set of managed disks generated taking into account the relationship between the managed disk and the resource group (314) determined and the resource group to which the managed disk is assigned .

  It is almost impossible to make intelligent data placement decisions without making a reasonable estimate of the workload required by the application, or at least those workloads. For example, if a user requests 100 GB of storage and a light performance is required, a single 100 GB logical disk may be allocated, but for high performance applications, a 10 disk array It may require the allocation of 10 10 GB logical disks throughout, and the striping of data across those logical disks. Unfortunately, most customers have no knowledge of how their workloads look.

  FIG. 4 illustrates a mechanism 400 for a user interface for obtaining a workload profile used by a virtual disk allocator when assigning a managed disk to a resource group according to an embodiment of the present invention. The workload profile describes characteristics of workload and desired throughput levels and may include information such as the amount of storage desired and allocation hints.

  Initially, a prepared workload profile may be provided (410). Referring to FIG. 2, the virtual disk allocator 272 may provide a workload profile prepared in advance in the memory 292. The workload profile 410 prepared in advance can be determined based on the characteristics of the customer environment in various industries and applications. By way of example, a well-known set of canned workloads may be provided, eg, SAP_OLTP, DB2 business intelligence, etc. With advice from an application expert, the customer first selects one of these given workloads 410.

  A workload profile may also be created automatically based on observations of customer workloads 412. Since each customer's workload has unique attributes, a better workload estimate can be obtained by observing storage access patterns in the customer's environment. Referring to FIG. 2, the virtual disk allocator 272 can base many of its decisions on the observed disk access behavior, and the virtual disk allocator 272 stores the observed disk access behavior in a database. In the memory 292 in the form of A user interface, such as a graphic interface that operates in conjunction with the virtual disk allocator 272, allows the user to group volumes, specify specific time frames, and then work based on the observed behavior of those volumes A load profile can be created. In this way, the virtual disk allocator 272 learns about the customer workload and improves its decision making over time. Note that the user interface may be a scripted application using a command line interface (CLI).

  The workload profile may also be provided by the intelligent software component 414. Referring to FIG. 2, virtual disk allocator 272 may also include an intelligent software component that provides a workload profile. These workload profiles may be based on special knowledge specific to the application.

  FIG. 5 illustrates a mechanism 500 for determining performance capabilities of a management disk or any resource group (hereinafter management disk). The managed disk performance capability is used by the virtual disk allocator 272 along with the workload profile to assign the managed disk to a resource group. Such mechanisms include any method ranging from manual input from the administrator to automated heuristics for deriving performance capabilities. Also included is a mechanism for determining performance by using calibration of capabilities through the use of a calibration workload in either a controlled or uncontrolled storage environment.

  One mechanism for determining managed disk performance capabilities may be the manual input approach 510. While this approach is simple, it is less desirable as an approach to understanding managed disk performance capabilities. For example, only a few managed disk performance profiles could be defined (eg, “mirrored disk”, “RAID-5 array with 8 disks”). Each of these profiles has specific default performance attributes. The administrator will then select the appropriate profile that matches the managed disk configuration. This approach will not take advantage of the great diversity of managed disk capabili.

  Another approach to ascertaining managed disk performance capabilities is to use managed disk configurator 512. A device-specific configurator could provide input regarding performance characteristics for a finite set of “understood” disk controllers and relationships between managed disks and storage controllers. For example, the FAStT900 configurator will provide a unified technique for creating RAID arrays, LUNs, and managed disks on a subsystem and will provide managed disk performance estimates for those managed disks. The configurator could also dynamically recognize the relationship between the management disk and the storage controller.

  In another approach, a controlled calibration 514 is used to run a specified I / O load on the managed disk while observing the behavior of the managed disk. Normally, controlled calibration is performed before space is allocated on the management disk, but may be performed later if space is reserved.

  In an uncontrolled calibration technique 516, the performance behavior of the managed disk is analyzed as the application runs a normal workload against the managed disk. If a managed disk is already allocated (and there is no free space reserved for calibration), this approach will be used. It is also useful to determine if the conditions have changed since the controlled calibration step (eg, the management disk is operating at degraded conditions). As an example, uncontrolled calibration may observe consistently high response times at a particular load point and estimate that this load point more accurately reflects the maximum throughput capability.

  Another approach uses a standardized interface 518 that can allow the SMIS specification to provide a conceptual model that identifies LUNs and the association of LUNs with physical resources, and the static performance capabilities of those elements.

  Allocation decisions can be made using other approaches similar to those described above. Similarly, a combination of the above approaches can be chosen. For example, the administrator runs a controlled calibration step for a particular type of first storage controller configured behind SANVC, and then for that same type of attribute for subsequent controllers, eg, “ These managed disks may look like those managed disks, "may be manually assigned.

  The workload parameters used by the virtual disk allocator 272 are based on their ability to accurately predict disk storage performance and based on their general availability through data collection tools. Selected. The workload parameters used are: maximum throughput on each managed disk, maximum random reads and writes per second, maximum mixed reads and writes per second, maximum sequential reads and writes per second, maximum mixed reads per second and Writes and latency at low and maximum loads are included.

  Mixed reads and writes are important in determining whether a component has bidirectional functionality. Short block random operations are used to determine processing power and cost, and large block sequential metrics are used to determine data transfer capacity and cost.

  These maximum throughputs are then used to determine the overhead per read and write operation and per read and write transfer megabyte. The overhead is then applied to the expected workload characteristics to determine the expected usage of the managed disk.

  Because managed disks have separate performance capabilities and have different performance capabilities when considering the performance capabilities of other resources associated with those managed disks, the virtual disk allocator 272 is A set of resource groups is defined, each resource group having an associated performance attribute for the set of managed disks belonging to that resource group. By considering the managed disk performance capabilities and the performance capabilities of their associated resources, the overall performance of the managed disk can be determined. These other resources may include physical disks or controllers that have management disks. However, it is not necessary to identify all possible resources associated with the managed disk. Instead, the virtual disk allocator 272 generalizes the association with resources when defining a resource group.

  Each resource group is given performance attributes (same as for managed disks). Resource group performance capabilities may be determined manually or through a controlled benchmark procedure called controlled calibration. Calibration can be performed on the identified managed disk at the time identified by the administrator. Calibration can be performed on the managed disk before the managed disk is assigned to the virtual disk. This may be included in the operating procedure that automates the creation of the management disk on the designated storage controller. Calibration will be performed on individual managed disks and all associated resource groups unless all managed disks in the group are assigned. The technique used in calibration is similar to that used in performing engineering measurements on competing disk subsystems.

  Since the managed disk can belong to more than one resource group, calibration of the resource group allows optimization of data allocation decisions. However, it is not necessary to find all potential relationships between managed disks and associated resources, and if no suspicious relationship exists, there will be no impact. Various implementations of the virtual disk allocator 272 may incorporate variable levels of sophistication when finding relationships.

  FIG. 6 illustrates a data structure 600 used by the virtual disk allocator 272 to extract key performance elements in a virtualized storage subsystem, according to an embodiment of the present invention. A data structure can be represented as a tree of nodes representing storage elements such as clusters, device adapters, individual disks or disk arrays, and any associated resources. However, those skilled in the art will appreciate that the present invention is not intended to be limited to the structure shown in FIG. Rather, a more general network of nodes may be used rather than a tree structure.

  Data structure 600 is used to provide a goal-oriented approach to storage allocation. As an example, SAN VC presents a model of storage that is slightly different from traditional storage controllers. The application server is presented aiming at a virtual disk corresponding to a logical disk in IBM Enterprise Storage Server (registered trademark) (ESS). Virtual disk allocator 272 identifies performance characteristics associated with a set of managed disks and resource groups, including throughput capabilities for a particular workload, and performance associated with the set of managed disks and resource groups. Create a virtual disk by estimating the resource utilization derived from the set of characteristics. The virtual disk is accessed, for example, via a SANV C primary node and the data for the virtual disk is cached in that node, which is a function corresponding to an ESS cluster. Referring to FIG. 6, storage for virtual disk 605 is allocated across one or more management disks 610, 620, 630 and 640. The illustrated managed disks are grouped into a managed disk group (MDG) 650. Virtual disk 605 is associated with a single MDG 650. In general, for maximum performance, virtual disk 605 will be striped across all managed disks in MDG 650.

  Each of the management disks 610, 620, 630 and 640 may be allocated as part of a single physical disk shared with other applications, or may be allocated as a whole RAID array of several disks. In addition, the storage administrator is given a great deal of flexibility in selecting managed disk configurations. As a result of associating with other applications, activity on one managed disk is relative to the performance of another managed disk, as if they are on the same storage device or behind the same storage controller, May have an impact. With reference to FIG. 6, the management disk 610 is associated with three resources 611, 612, and 613, while the management disk 620 is associated with resources 613, 621, and 622. Since resource 613 is associated with both managed disks 610, 620, each of the managed disk performance characteristics may be unknown and highly variable. Thus, the key to intelligent data placement decisions is to obtain and maintain a reasonable performance estimate regarding the capabilities and behavior of the managed disk.

  According to an embodiment of the invention, each managed disk is associated with one or more resource groups (RGs). RG1 (660), RG2 (670), and RG3 (680) are defined by virtual disk allocator 272. The RG1 (660) includes a management disk 610 and resources 611, 612, and 613. RG2 (670) includes management disks 620 and 630 and a part of the management disk 640, and has resources 613, 621, 622, 631, 632, and 641. RG3 (680) includes a management disk 640 and a portion of resources 632 and 641. As can be seen, resource 613 is shared by management disks 610 and 620, and resources 632 and 641 are shared by management disks 630 and 640. Because managed disks and associated resources are grouped to allow observation of interactions within and between resource groups, enhanced performance estimates for managed disk capabilities and behavior are provided for virtual disks. Can be done by allocator 272

  For example, during the volume selection process, the virtual disk allocator 272 constantly monitors the usage of managed disks and resource groups, as well as how to maintain their usage for the components of the configuration tree in the VPA for ESS. To do. When selecting a potential managed disk for reliance, the impact is evaluated for all of the resource groups to which a managed disk belongs. The utilization of the management disk is then assumed to be the highest among all related RGs. This is similar to the technique of navigating the configuration tree used by the current version of VPA. Virtual disk allocator 272 recommends storage allocations that meet specified capacity and performance requirements.

  Referring back to FIG. 2, the virtual disk allocator 272 is improved by knowing how the storage element is actually operating, but does not rely on very accurate information, which is This is because the allocator 272 can operate for heterogeneous types of storage from different vendors. However, accurate real-time or historical performance data separates one vendor's product from others as well as biases storage allocation and keeps it away from potentially competing workloads for the time period covered Can be used for

  An important aspect of virtual disk allocator 272 involves the use of capacity and performance structures to balance storage allocation across available resources. If multiple selections are possible in the virtual disk allocator 272, capacity and performance structures can be used to bias the allocation to a set of resources using pseudo-random numbers. Several sample assignments can be selected in this way, and the best of the samples is chosen as the answer. This approach prevents the algorithm from making the same recommendation in similar situations and increases the spread of the workload across the available resources. In this way, storage allocation is biased towards the elements in the network that are most capable of handling the specified workload.

  FIG. 7 illustrates a flowchart 700 of a method for managing storage allocation in a virtual storage system, according to an embodiment of the present invention. The method includes obtaining 710 a user-defined workload profile and determining 720 performance characteristics of the managed disk. A relationship between the managed disk and the resource group is determined based on user-defined or automated input (730). A virtual disk including a set of managed disks is created (740) taking into account the resource group to which the managed disk is assigned.

  The process illustrated in connection with FIG. 7 can include one or more of a computer readable medium or carrier, for example, one or more of the fixed and / or removable data storage devices 288 illustrated in FIG. Alternatively, it may be embodied in other data storage devices or data communication devices. Data storage device 288 or computer program 290 may be loaded into memory 292 to configure administrator 270 or virtual disk allocator 272 for execution. Such computer programs include instructions that, when read by a processor such as processor 294 of FIG. 2, perform the steps or elements of the present invention to administrator 270 or virtual disk allocator 272. Make the steps necessary to do it.

  The foregoing descriptions of exemplary embodiments of the present invention have been presented for purposes of illustration and description. It is not intended to be exhaustive or to limit the invention to the precise form disclosed. Many modifications and variations are possible in light of the above teaching. It is intended that the scope of the invention be limited not by this detailed description, but rather by the claims appended hereto.

1 illustrates a computer network in the form of a local area network. 1 illustrates one embodiment of a storage area network according to an embodiment of the invention. Fig. 4 illustrates a table of attributes incorporated into a virtual disk allocator according to embodiments of the present invention. FIG. 4 illustrates a mechanism for a user interface to obtain a workload profile that is used by a virtual disk allocator in assigning a managed disk to a resource group according to an embodiment of the present invention. Fig. 4 illustrates a mechanism for determining performance capabilities of a managed disk according to one embodiment of the invention. Fig. 4 illustrates a data structure used by a virtual disk allocator to extract important performance factors in a virtualized storage subsystem, according to an embodiment of the present invention. 2 illustrates a flow diagram of a method for managing storage allocation in a virtual storage system, according to an embodiment of the invention.

Claims (44)

A program storage device,
Including program instructions executable by a processing device that performs operations for managing storage allocation in a virtual storage system, the operations comprising:
Evaluating a workload profile;
Determining the performance characteristics of the managed disk;
Determining a relationship between the managed disk and a resource group;
Recommending allocation of at least one virtual disk comprising the set of managed disks based on a resource group to which the managed disk is allocated.
Program storage device. The operation is
Further comprising defining the workload profile;
The program storage device of claim 1. The operation is
Creating a virtual disk containing a set of managed disks based on a resource group to which the managed disk is assigned;
The program storage device according to claim 1 or 2.   4. A program storage device according to claim 1, 2 or 3, wherein said step of determining a relationship is based on user input.   The program storage device of any one of claims 1 to 3, wherein the step of determining a relationship is based on automatically provided input.   4. The program of claim 3, wherein the step of creating a virtual disk further comprises creating a plurality of virtual disks in the set of managed disks based on the resource group to which the managed disk is assigned. Storage device. Selecting a location for the virtual disk by identifying a set of performance characteristics associated with a set of managed disks and a resource group that includes throughput capabilities for a particular workload; and
Further comprising estimating optimal resource utilization based on a set of performance disks associated with a set of managed disks and a resource group;
A program storage device according to any one of the preceding claims.   The step of defining a workload profile further comprises selecting at least one selected from the group consisting of a desired storage amount, an identification of the workload characteristics and a desired throughput level, and an allocation hint. Item 3. The program storage device according to Item 2.   A program storage device according to any preceding claim, further comprising the step of obtaining performance capabilities of the management disk.   The program storage device of claim 9, wherein the step of obtaining managed disk performance capabilities further comprises providing manual input indicative of managed disk performance capabilities.   The program storage device of claim 9, wherein obtaining the performance capability of a managed disk further comprises receiving a performance capability from a managed disk configurator.   The step of receiving performance capabilities from a managed disk configurator further comprises obtaining information about the performance characteristics of managed disks and their associated resource groups, the resource group comprising a potential between managed disks. The program storage device of claim 11, wherein the program storage device identifies a general relationship.   The program storage device of claim 9, wherein obtaining the performance capability of a managed disk further comprises performing a controlled calibration and using the result from the controlled calibration.   14. The program of claim 13, wherein the step of performing controlled calibration comprises running a specified I / O load for the management disk to identify the behavior of the management disk. Storage device.   The program storage device of claim 9, wherein obtaining the performance capabilities of a managed disk further comprises performing an uncontrolled calibration and using the result from the uncontrolled calibration.   16. The program of claim 15, wherein the step of performing uncontrolled calibration comprises analyzing the performance behavior when an application operates as an uncontrolled workload to identify the behavior of the managed disk. Storage device.   10. The program storage device of claim 9, wherein the step of obtaining managed disk performance capabilities further comprises receiving input from a user via an interface.   A program storage device according to any preceding claim, wherein the step of determining performance characteristics of a managed disk further comprises identifying a resource group to which the managed disk belongs.   The program storage device of claim 17, wherein the user input includes a selection from a managed disk performance profile.   The program storage device of claim 17, wherein the user input includes a selection from a performance profile resource group. Evaluating a workload profile;
Determining the performance characteristics of the managed disk;
Determining a relationship between the managed disk and a resource group;
Recommending allocation of at least one virtual disk comprising the set of managed disks based on a resource group to which the managed disk is allocated.
Method. Further comprising defining the workload profile;
The method of claim 21. Creating a virtual disk containing the set of managed disks based on a resource group to which the managed disk is assigned;
The method according to claim 21 or 22.   24. The method of claim 21, 22 or 23, wherein the step of determining a relationship is based on user input.   24. A method according to any one of claims 21 to 23, wherein the step of determining a relationship is based on automatically provided input.   24. The method of claim 23, wherein the step of creating a virtual disk further comprises creating a plurality of virtual disks in the set of managed disks based on the resource group to which the managed disk is assigned. Selecting a location for the virtual disk by identifying a set of performance characteristics associated with a set of managed disks and a resource group that includes throughput capabilities for a particular workload; and
Further comprising estimating optimal resource utilization based on a set of performance disks associated with a set of managed disks and a resource group;
27. A method according to any one of claims 21 to 26.   Said step of defining a workload profile further comprises selecting at least one selected from said group of desired storage amount, identification of said workload characteristics and desired throughput level, and allocation hints; The method of claim 22.   29. A method according to any one of claims 21 to 28, further comprising the step of obtaining performance capabilities of the managed disk.   30. The method of claim 29, wherein the step of obtaining managed disk performance capabilities further comprises providing manual input indicating managed disk performance capabilities.   30. The method of claim 29, wherein obtaining the performance capability of a managed disk further comprises receiving a performance capability from a managed disk configurator.   The step of receiving performance capabilities from a managed disk configurator further comprises obtaining information about the performance characteristics of managed disks and their associated resource groups, the resource group comprising a potential between managed disks. 32. The method of claim 31, wherein a generic relationship is identified.   30. The method of claim 29, wherein obtaining the performance capability of a managed disk further comprises performing a controlled calibration and using the result from the controlled calibration.   34. The method of claim 33, wherein the step of performing a controlled calibration comprises operating a designated I / O load for the management disk to identify the behavior of the management disk.   30. The method of claim 29, wherein obtaining the managed disk performance capabilities further comprises performing an uncontrolled calibration and using the result from the uncontrolled calibration.   36. The method of claim 35, wherein performing the uncontrolled calibration comprises analyzing the performance behavior when an application operates as an uncontrolled workload to identify the behavior of the managed disk.   30. The method of claim 29, wherein obtaining a managed disk performance capability further comprises receiving input from a user via an interface.   38. A method according to any one of claims 21 to 37, wherein the step of determining performance characteristics of a managed disk further comprises identifying a resource group to which the managed disk belongs.   38. The method of claim 37, wherein the user input comprises a selection from a managed disk performance profile.   38. The method of claim 37, wherein the user input includes a selection from a performance profile resource group.   41. A volume provisioning advisor for managing storage allocation in a virtual storage system, the volume provisioning advisor being operable to perform the method of any one of claims 21-40 The volume provisioning advisor.   42. The volume provisioning advisor of claim 41, wherein the resource group includes a name, performance attributes, and a set of managed disks belonging to the resource group.   43. A volume provisioning advisor according to claim 41 or 42, wherein a managed disk belongs to one or more resource groups, and the resource group identifies potential relationships between managed disks.   41. A computer program comprising program code means adapted to perform the method of any one of claims 21 to 40 when said program is executed on a computer.

Images